Short- and Lon -term Assessment of Heart Rate srariabilitv for Risk Stratification After Acute Myocardial Infarction Lii Fei, MD, PhD, Xavier Copie, MD, Marek Malik, PhD, MD, and A. John Camm, MD i Depressed heart mte variability (HRV) has been shown to be a powerful and independent risk factor in patients following acute myacardial infarction (AM). A detailed comparison of the predictive values between short- and long-term HRV has not been made. The predictive value of short-term HRV for 1-year total cardiac mortality was studied in 700 consecutive patients after AM. All pa- tients underwent 24-hour Holter monitoring before dis- charge from the hospital (5 to 8 days after AMI) and were followed up for 1 year. Short-term HRV was com- puted as the standard deviation of all normal RR inter- vals (SDNN) from a 5minute stationary period selected from 24-hour Holter electrocardiographic recordings. long-term HRV was computed as an HRV index over the entire 24 hours. There was a significant but relatively poor correlation between SDNN and HRV index (r = 0.5 1, p <O.OOl ). The positive predictive accuracy of SDNN for 1-year mortality ( 13% to 18%) was lower than the HRV index (17% to 43%) over a mnge of sensitivity of 25% to 75%. Assessment of HRV index in ~35% of the patients preselected by the lowest SDNN was able to achieve predictive power similar to that of HRV index assessed in all the patients. These data suggest that lower predischarge short-term HRV is associated with increased 1-year total cardiac mortality in patients after AM. Analysis of long-term HRV for postinfarction risk stratification can safely be limited to patients preselected by depressed short-term HRV measures. (Am J Cardiol 1996;77:68 l-684 A nalysis of heart rate variability (HRV ) has pro- vided a noninvasive method for assessing au- tonomic influence on the heart. HRV has been shown to be depressed in various clinical settings.‘,* Several investigators 3-6 have demonstrated that depressed HRV is a powerful risk predictor in patients after acute myocardial infarction (AMI), independent of other established risk factors. The predictive value of HRV has almost exclusively been derived from data analyzed from 24-hour Holter electrocardio- grams. However, there are technical difficulties and cost implications that limit the assessment of HRV from ambulatory 24-hour Holter recordings. Short- term recordings are obviously more practical for the clinical application of HRV assessment. However, data on short-term analysis of HRV for postinfarc- tion risk stratification are scarce.7*8 Bigger et al8 re- ported that spectral HRV from short-term recordings (2 to 15 minutes) randomly selected from 24-hour Holter electrocardiograms strongly predicts postin- farction mortality. However, the predictive power of short-term HRV (positive predictive accuracy up to 31%)* is lower than that of HRV calculated from 24-hour recordings (positive predictive accuracy up to 41%) .9 Whether short-term HRV can be used for screening patients for further risk assessment is un- known. The cost-effectiveness of short- and long- From the Department of Cardiological Sciences, St. George’s Has- pital Medical School, Cranmer Terrace, London, United Kingdom. Manuscript received August 7, 1995; revised manuscript received and accepted November 7, 1995. Address for reprints: Lti Fei, MD, PhD, Krannert Institute of Cardi- ology, 1 1 1 1 West 10th Street, Indianapolis, Indiana 46202- 4800. term assessments of HRV for postinfarction risk stratification remains to be fully defined. This study was specifically designed to assess the predictive value of short- and long-term HRV based on data from St. George’s Post Infarction Survey Program. METHODS Patients: Seven hundred patients with documented AM1 admitted to St. George’s Hospital were enrolled in this study. The definition of infarction, the enroll- ment criteria, data acquisition techniques, and fob low-up have previously been published.5 All patients underwent a risk stratification protocol before hos- pital discharge between days 5 and 8 following AMI. This included a symptom-limited treadmill exercise test (Bruce protocol), signal-averaged electrocar- diography, and 24-hour Holter monitoring. Left ven- tricular ejection fraction was assessedby either car- diac catheterization (exercise test positive) or radionuclide-gated blood pool scanning (exercise test negative). The clinical characteristics of the 700 patients are summarized in Table I. All patients were followed up for at least 1 year. Total cardiac mor- tality during the first year after AM1 was used as the end point in this study. Analysis of heart rate variability: Short-term HRV was computed as the SD of all normal sinus RR in- tervals (SDNN) from a short (5 to 20 minutes) clean and stable period of the 24-hour electrocardiographic recording. This short period was selected by the fol- lowing criteria: ( 1) sinus rhythm with no ectopic beats or artifacts (clean), and (2) none of the RR intervals during the entire period differed from the initial RR interval by >20% (stationarity ) . The first CORONARY ARTERY DISEASE/HEART RATE VARIABILITY AFTER MYOCARDIAL INFARCTION 681